The Dialogical Brain Contributions of Emotional Neurobiology to Understanding the Dialogical Self Marc D. Lewis University of Toronto Abstract. Using theory and data from emotional neurobiology, I suggest a neurally realistic model of Hermans’ dialogical self. The model is premised on Hermans’ idea of voicing and its implications for motivation, action and subjectivity. Because states of motivated attention unify brain activity, coexisting I-positions are as problematic for neuroscience as they are for psychology. To overcome this problem, I postulate an internal monologue in which the familiar I-position is subserved by an attentional system in the orbitofrontal cortex, linked with nearby affective and premotor areas. This internal monologue is fueled by gist-like perceptual expectancies of an- other’s response, and it perpetuates and adjusts itself by updating these expectancies. A second I-position may be underpinned by an attentional system in the anterior cingulate cortex and its connections. These two attentional systems are partly independent, and they compete for control based on changes in emotional content and intensity. Thus, switching activation between them may account for semi-autonomous, but not coexisting, I-positions. Key Words: brain, emotion, expectancy, internal dialogue, subjectivity Mary is alone in the kitchen, cooking for tonight’s guests. She goes about the routine tasks of chopping and mixing, with little awareness of anything except a vague pleasure in familiar actions and anticipation of a tasty outcome. Then she notices that she put the stove on too high, and the rice will soon be ruined. There is not enough time to start the rice again if all is to be ready when the guests arrive. She notices a change in her mood, a loss of pleasure and increase in anxiety, but this is expectable under the circumstances. Then, as she rushes around trying to repair the situation, she decides to stop and ‘look inward’, as the yoga teacher suggests, and notices that there are phrases in her mind: ‘What do you expect . I can’t do everything on my own. It’s not my fault.’ And she notices that these phrases are directed at someone, but she isn’t sure who it is. It seems a bit like her mother, who is about to arrive, or her husband, who does not appreciate how Theory & Psychology Copyright © 2002 Sage Publications. Vol. 12(2): 175–190 [0959-3543(200204)12:2;175–190;022628] 176 THEORY & PSYCHOLOGY 12(2) difficult it is to cook a perfect meal. But the person she is addressing is apparently much closer, because that person now responds sarcastically, ‘Oh, and whose fault is it?’ How can the self be one yet many? This question has preoccupied philosophers and writers for centuries. It is a question that bridges worlds as different as Buddhist meditation, psychoanalysis and cognitive science (Varela, Thompson, & Rosch, 1991), it underlies our fears about personality changes due to strokes and pharmacology, and it guides theories of psycho- diagnostics and psychotherapy that try to make sense of self-induced suffering. Recently, social and personality psychologists have attempted to address the multiplicity of the self by replacing terms such as ego and superego with the more contemporary language of self schemas and narrat- ive structure. However, these conceptualizations essentially exchange one set of metaphors for another, and they propose static mechanisms for explaining what appears to be a very active process (Hermans, 1996). To move beyond this impasse, Hermans models the multiplicity of the self in terms of voiced positions engaged in dialogue. This formulation is clearly action-based, and it captures some of the phenomenology revealed by clinical reports. But is it correct? Does it point toward tangible psychological mechanisms, or merely provide another set of metaphors? Psychologists are increasingly looking toward the brain in order to ground their modeling in biological reality, and the discipline of cognitive neuro- science has compiled a great deal of data to assist them. Using these data, I suggest a neuroscientific instantiation of Hermans’ model of the dialogical self in order to evaluate its core tenets, increase its precision and help extract it from its remaining metaphors. Following Hermans, I see the agentic notion of voicing as an important refinement, and I use it as a bridge to the brain. I discuss the brain’s premotor systems and the attentional states that guide them, and I show how these and related systems are rooted in emotion and anticipation. I then speculate on how these systems might be engaged in producing internalized voices and responding to those voices as if they came from someone else. Finally, I address the claim that different voices or positions coexist, occupying the same mental space at the same time. This portrayal of multiplicity creates problems for neurobiology just as it does for psychology. Yet they are creative problems that point toward a fundament- ally new perspective. Voicing in the Dialogical Self: Embodiment, Agency and Position In order to move beyond the metaphors of previous accounts, including the modern metaphors of information processing and narrative, Hermans (1996) highlights the active voice in the multiple self. The ‘agentlike qualities of the LEWIS: THE DIALOGICAL BRAIN 177 I’ (p. 42) give each of the self’s various positions a vital, active part to play in the internal dialogue. In fact, positioning assumes the status of a verb to convey this activity, and this allows Hermans to break away from the more passive constructs of schema, script and story-line. Hermans frames the dialogical self as an interaction between various voices. These I-positions take turns in an internal dialogue, like interacting characters in a story. They agree or disagree and tell stories from their own perspectives. Moreover, they have the capacity to change or evolve by taking into consideration the perspective of the other. This means that they must hear as well as speak, and indeed the concept of voice is meaningless unless it is related to the perception of other voices. For Hermans, ‘voice assumes an embodied actor located in space together with other actors’ (p. 44). Hermans’ focus on embodiment places him in league with contemporary cognitive scientists for whom information processing is the most recent in a line of inadequate metaphors. The information-processing account, which has been the standard language of cognitive psychology for years, compares cognition to the sequential analysis of symbols by a digital computer. Despite its enormous contribution, this metaphor has lost utility because it cannot be squared with subjectivity, intentionality and motivation, qualities that set humans apart from machines. In its place, theorists have proposed cognitive processes that are fundamentally embodied. First, they are viewed as emergent, self-organizing, global gestalts that arise from reciprocal interactions among processing units, not linear sequences (Clark, 1996; Varela et al., 1991). Second, they are necessarily affective or motivated, linking biological requirements to the formation and manipulation of societal meanings (Fogel, 1993; Freeman, 1995; Lewis, 1995). Indeed, Hermans and Kempen (1993) provide a central place for emotion in their model, and Hermans and Hermans-Jansen (in press) propose that changes in voiced positions can be explained as phase transitions in non-linear dynamic systems. Thus, their embodied account of the multiple self fits with an emerging zeitgeist in cognitive science. In cognitive science at large, the move toward embodiment includes a commitment to understand the brain as the basis of cognition. The richly distributed, reciprocally interactive and self-organizing character of neural activity provides a radical alternative to the linear sequences of symbol- processing machines (e.g. Thelen & Smith, 1994; Varela et al., 1991). It follows that Hermans’ move toward embodiment should be compatible with neural realism. Moreover, his emphasis on voice provides a useful entry point to test this compatibility. Voicing, construed as action, points toward the brain regions and subsystems directly involved in planning and generat- ing voluntary speech. Voicing, construed as listening, points toward the attentional systems that anticipate others’ speech and prepare for one’s own. Thus, a good place to look for dialogicity in the brain is in systems where attention and action are integrated. These are generally acknowledged to be 178 THEORY & PSYCHOLOGY 12(2) the frontal and prefrontal cortical systems, and they are held responsible for a focused sense of the self acting in the world. By studying the character of these systems, their dependence on emotion and their contribution to learning and memory, we can speculate on how a dialogical self might actually be housed in a dialogical brain. One of the most difficult and intriguing question for the dialogical self is how one can be both subject and object in the same dialogue, as was the case with Mary. To address this question, this article is restricted to discussion of ‘internal’ dialogues, that is, dialogues between one’s various I-positions (Hermans & Hermans-Jansen, in press). This exploration is guided by a contention of Hermans’ that is particularly challenging from a neural perspective. Hermans (1996) emphasizes that the positions in the dialogical self are distributed in an imaginal space. This leads him to propose the ‘simultaneous existence’ (p. 46) of internalized voices. However, as we shall see, attention, action and motivation appear to synchronize the brain, such that numerous subsystems become highly coordinated in real time. It is difficult to imagine how semi-autonomous positions could coexist at the same moment in such a unitary brain, and this problem indeed recapitulates the classic difficulty of seeing the self as both unitary and multiple. Thus, what is problematic for a neural explanation is problematic for the study of the self more broadly: the reconciliation of unity and multiplicity. We now look to the brain to see how this challenge can be addressed.